It seems that all articles dealing with cheating must start out by saying that the phenomenon has not been studied before. Similarly, it would appear that all researchers working with trust have made a pact to always begin by emphazising the difficult and elusive nature of trust. Trust, or so they claim, is extraordinarily hard to define and particularly multifaceted.
I think there is some mystification going on here. But the confusion is genuine when it comes to whether artifacts or arrangements (such as contracts, police forces etc.) can be said to eliminate the need for trust or should be seen as features which in fact increase trust (in most cases).
The former position (the one most often found in sociological accounts as opposed to economic ones) hinges on the idea that trusting means putting one at the mercy of someone else. But I think there are two problems here:
1) We are led to assume that people trust indiscriminately. But people base their trust on all sorts of signals even they don’t carry around actual contracts.
2) A contract is in fact a way of putting yourselves at the mercy of someone else (you specifically make that person able to sue you should you fail to uphold your part of the bargain). A contract merely makes the mercy aspect work both ways at the same time.
So, I think the latter version makes more sense. If you and I conduct business through a bank then we (can) trust each other even if the bank plays a role in securing that trust.